vit-base-patch16-224-ve-U13b-80R

This model is a fine-tuned version of google/vit-base-patch16-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4109
• Accuracy: 0.8913

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5.5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3158	0.99	51	1.2967	0.3478
1.0472	2.0	103	0.9694	0.5
0.6641	2.99	154	0.7911	0.7391
0.5615	4.0	206	0.6850	0.7391
0.3458	4.99	257	0.4109	0.8913
0.3156	6.0	309	0.5213	0.8043
0.141	6.99	360	0.4793	0.8478
0.2016	8.0	412	0.6031	0.7826
0.2444	8.99	463	0.7324	0.8043
0.1501	10.0	515	0.6392	0.8043
0.1256	10.99	566	0.9706	0.7826
0.2421	12.0	618	0.8059	0.7826
0.103	12.99	669	0.7601	0.8478
0.1353	14.0	721	1.1986	0.7391
0.1095	14.99	772	1.0279	0.7609
0.065	16.0	824	1.2043	0.6957
0.1777	16.99	875	0.9779	0.8043
0.0813	18.0	927	1.3356	0.7391
0.2552	18.99	978	0.8483	0.8261
0.0941	20.0	1030	0.7106	0.8696
0.0486	20.99	1081	0.8359	0.8261
0.0361	22.0	1133	0.8710	0.8261
0.0361	22.99	1184	1.0301	0.8043
0.0136	24.0	1236	0.9015	0.8261
0.1441	24.99	1287	0.9958	0.8043
0.0181	26.0	1339	1.0793	0.7826
0.0612	26.99	1390	0.9678	0.8043
0.0814	28.0	1442	1.0320	0.7826
0.0479	28.99	1493	1.1845	0.7826
0.06	30.0	1545	1.2026	0.7826
0.0777	30.99	1596	1.1574	0.7826
0.0747	32.0	1648	1.3104	0.7609
0.0181	32.99	1699	1.1145	0.8043
0.0652	34.0	1751	1.1691	0.8043
0.0242	34.99	1802	1.2415	0.8043
0.0043	36.0	1854	1.1841	0.7826
0.0318	36.99	1905	1.2475	0.8043
0.0092	38.0	1957	1.2452	0.8043
0.0194	38.99	2008	1.2395	0.8043
0.0376	39.61	2040	1.2345	0.8043

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0